MR utilizes genetic variants strongly associated with an exposure variable as an ‘instrument’ to test for causal effects of the exposure on an outcome variable. This approach minimizes the risk of spurious findings due to confounding or reverse causation present in observational studies, provided that the following assumptions are met: 1) the genetic instrument is predictive of the exposure variable, 2) the genetic instrument is independent of confounders, and 3) the genetic instrument is not directly associated with the outcome variable, other than by its potential causal effect through the exposure (i.e. there is no directional pleiotropy)65. Two-sample MR refers to the application of MR methods to well-powered summary association results estimated in non-overlapping sets of individuals22 in order to reduce instrument bias towards the exposure-outcome estimate.
